1. Introduction
Although transposons have been identified in numerous species as DNA segments with particular structures, in only a few cases has it been possible to detect phenotypic consequences of their activity. Such consequences are particularly evident in the case of plant transposons which, when inserted in pigmentation genes, give rise to unstable mutant alleles conferring variegated colour phenotypes (Bonas et al., Reference Bonas, Sommer and Saedler1984; Fedoroff et al., Reference Fedoroff, Furtek and Nelson1984; Kroon et al., Reference Kroon, Souer, de Graaff, Xue, Mol and Koes1994; Inagaki et al., Reference Inagaki, Hisatomi, Suzuki, Kasahara and Iida1994; Chopra et al., Reference Chopra, Brendel, Zhang, Axtell and Peterson1999). In maize, Antirrhinum and Petunia, detailed analysis of alleles of this type has led to the use of transposons as tools for systematically generating mutants (Carpenter & Coen, Reference Carpenter and Coen1990; Luo et al., Reference Luo, Coen, Doyle and Carpenter1991; van Houwelingen et al., Reference May, Liu, Vollbrecht, Senior, Rabinowicz, Roh, Pan, Stein, Freeling, Alexander and Martienssen1998; May & Martienssen, Reference May and Martienssen2003). An advantage of transposon mutagenesis is that it offers the prospect of isolating genes from unstable alleles or, conversely, selecting for mutants carrying transposon insertions or transposon-induced deletions in known genes (Koes et al., Reference Koes, Souer, van Houwelingen, Mur, Spelt, Quatrocchio, Wing, Oppedijk, Ahmed, Maes, Gerats, Hoogeveen, Meesters, Kloos and Mol1995; Ingram et al., Reference Ingram, Doyle, Carpenter, Schultz, Simon and Coen1997; Davies et al., Reference Davies, Motte, Keck, Saedler, Sommer and Schwarz-Sommer1999; Galego & Almeida, Reference Galego and Almeida2002; May et al., Reference May, Liu, Vollbrecht, Senior, Rabinowicz, Roh, Pan, Stein, Freeling, Alexander and Martienssen2003). Furthermore, transposition can in some instances be exploited to genetically mark clones of cells, hence for lineage analysis or for studying cell autonomy in gene action (Dawe & Freeling, Reference Dawe and Freeling1990; Dudley & Poethig, Reference Dudley and Poethig1993; Carpenter & Coen, Reference Carpenter and Coen1995; Hantke et al., Reference Hantke, Carpenter and Coen1995; Vincent et al., Reference Vincent, Carpenter and Coen1995; Rolland-Lagan et al., Reference Rolland-Lagan, Bangham and Coen2003).
The possibilities above have been exploited for detailed studies on the evolution of genetic and developmental mechanisms incorporating comparisons of mutants in diverse, distantly related, model species. For example, site-directed selection for transposon insertions has been used to investigate the role of Antirrhinum counterparts of the Arabidopsis Ap2 gene in flower development (Keck et al., Reference Keck, McSteen, Carpenter and Coen2003). Equivalent tools are, however, unavailable for comparisons between Antirrhinum and its close relatives. Thus, as an exploratory step towards expanding transposon mutagenesis to a genus closely related to Antirrhinum, we have investigated the genetic molecular basis of a case of flower colour variegation in Linaria, a member of the Antirrhinum tribe (Olmstead et al., Reference Olmstead, dePamphilis, Wolfe, Young, Elisens and Reeves2001; Vargas et al., Reference Vargas, Rossello, Oyama and Güemes2004). We show that such variegation results from activity of a CACTA transposon inserted in the gene encoding dihydroflavonol-4-reductase and that Tl1 transposes in cells giving rise to gametes.
2. Materials and methods
(i) Origin of plants
The Linaria plants considered here (Fig. 1) were found in a garden population of unknown origin. Many morphological and colour variants occurred in the population, which could possibly comprise more than one species or species hybrids. Assigning individuals to particular species was therefore problematic. One possibility was that plants with variegated colour belonged to or comprised part of Linaria maroccanna, a species from which popular garden varieties are derived. These plants had reniform seed with distinct transverse ridges and a style with lateral lobes bearing discrete stigmatic areas, all traits common to L. maroccanna and a few other species (Sutton, Reference Sutton1988). However, it was unclear whether capsules had conspicuously unequal loculi, a trait which, together with those above, is common to only L. maroccana and L. gharbensis.
(ii) Primers
The following primers were used:
i1: AA(A/G)(C/A)GNCA(C/T)ACNGA(C/T)CCNGGNACNAT(A/C/T)AC
i2: TTNACNGC(T/C)TG(G/A)TG(G/A)TCNGC(G/A)TT(T/C)TT
p1: GGNTT(C/T)AT(A/C/T)GGN(A/T)(G/C)NTGG(T/C)TNGTNATG
p2: AA(G/A)ATGACNGGNTGGATGTA(C/T)TT
p3: GGTCACTGGAATCGGTTTCATCG
p4: GTGGCTATACCATTCGTGCAAC
p5: CTCATGACCAGCCATGAGCCAA
p6: CTGCCAACGTGACGAACTTCAC
p7: CCACAATAACATGTCCGTATC
p8: TGCAACTCACAATTAGCT
p9: CCGTTGGCTCCAGTAACGCAT
p10: GCGATACTAGGCAACAAGTTC
Tam1-3′: CCAGAGCCCAATTTTGTTGTAGTG
The positions and orientations of primers p1 to p9 are indicated with arrowheads in Fig. 2C.
(iii) Molecular analysis
cDNA Segments of LPAL and LINC were isolated by RT-PCR using RNA extracted from red inflorescences and degenerate primers designed according to aligned sequences of genes encoding dihydroflavonol-4-reductase (p1 and p2; 1 and 2 in Fig. 2C) and flavonol-3-hydroxylase (i1 and i2) from various species. A genomic segment containing the right end of Tl1 and its flanking Lpal sequence was isolated by PCR amplification of DNA from Lpal-1 homozygotes using an LPAL-specific primer, p3, and a primer for the 3′ end of the Antirrhinum transposon Tam1 (Tam1–3′) at low annealing temperature (55°C). A genomic segment comprising 478 bp of the right end of Tl1 and its flanking Lpal-2 region was then obtained by inverse PCR using p4 and p5 on DNA cut with Sau3A prior to ligation. Using the same approach on DNA from fully red revertants resulted in amplification of a segment spanning the site of Tl1 excision and a further 453 bp to its left. A segment containing 342 bp of the left end of Tl1 was then obtained by inverse PCR using p6 and p7 on DNA from Lpal-1 homozygotes cut with Sau3A prior to ligation. The LPAL transcription start site was determined by 5′RACE using p3 and p5. A nearly full length cDNA was then obtained by 3′-RACE using p8 as specific primer. A genomic segment spanning the transcribed region was amplified using p8 and p10. Segments described above were cloned in pBluescript SK (Stratagene) and sequenced automatically (Stabvida, Oeiras). Segments from reversion alleles, carrying Tl1 excision footprints, were obtained by PCR using p5 and p6. Sequences of these segments were then determined directly on PCR products using p9.
Sequences for LPAL, Tl1 and LINC have been deposited in the GenBank Data Library under accession nos. EF517133, EF517134 and EF517135, respectively.
3. Results and discussion
(i) Germinal instability of flower colour
Plants with variegated colour had ivory flowers and green stems with clonal patches of red and purple cells respectively, a phenotype very similar to those conferred by some unstable alleles of pigmentation genes in Antirrhinum (Coen & Carpenter, Reference Coen and Carpenter1986; Schwarz-Sommer et al., Reference Schwarz-Sommer, Davies and Hudson2003; Fig. 1). Typically, insertion of a transposon in a pigmentation gene blocks gene expression giving non-pigmented tissue. However, the transposon is occasionally excised from the gene in dividing cells, hence giving clones of pigmented cells on the non-pigmented background. In addition, the transposon can excise in cells giving rise to gametes and therefore a few plants with wild-type phenotype (revertants) can be recovered in progenies of homozygotes for the unstable allele. To test whether the apparent instability in Linaria occurred in germinal tissue we examined pigmentation phenotypes in progenies from crosses between plants with variegated colour (these plants were self-incompatible). In nearly 700 progeny from five crosses, 10 plants had fully red flowers and purple stems whereas the remaining ones had the parental phenotype. Thus, by analogy to Antirrhinum, an explanation for the variegated colour phenotype and genetic behaviour of Linaria plants was that they carried an active transposon in a pigmentation gene.
(ii) Lpal-1: an unstable mutant allele of the gene encoding dihydroflavonol-4-reductase
In Antirrhinum, red colour depends on structural genes encoding enzymes required for anthocyanin synthesis, and on regulatory genes that control the expression of structural genes (Almeida et al., Reference Almeida, Carpenter, Robbins, Martin and Coen1989; Goodrich et al., Reference Goodrich, Carpenter and Coen1992; Schwinn et al., Reference Schwinn, Venail, Shang, Mackay, Alm, Butelli, Oyama, Bailey, Davies and Martin2006). Null mutants for structural genes are in general devoid of anthocyanin pigment in all plant parts whereas regulatory mutants show altered colour patterns relative to wild-type. In Linaria, except for the clonal patches of pigmented cells, all plant parts lacked red pigment, suggestive of mutation in a structural gene. Null mutations affecting enzymes required at early steps of the anthocyanin synthesis pathway in Antirrhinum, such as chalcone synthase, result in albino flowers whereas mutations affecting late steps give ivory flower colour identical to that in Linaria (see Coen et al., Reference Coen, Robbins, Almeida, Hudson, Carpenter, Berg and Howe1989). Late steps include those requiring flavonol-3-hydroxylase and dihydroflavonol-4-reductase, the products of the INCOLORATA (INC) and PALLIDA (PAL) genes respectively (Martin et al., Reference Martin, Carpenter, Sommer, Saedler and Coen1985, Reference Martin, Prescott, Mackay, Bartlett and Vrijlandt1991; Coen et al., Reference Coen, Carpenter and Martin1986).
To determine whether variegated plants were mutant at the counterparts of either INC or PAL (LINC or LPAL) we probed Northern blots of inflorescence RNA from variegated and fully red sibs derived from the crosses described above with cDNA segments for the LINC or LPAL genes (Fig. 2A). LINC mRNA was detected in both variegated and red inflorescences whereas LPAL mRNA was detected in red but not in variegated inflorescences, suggesting that mutation at Lpal caused colour variegation. Consistent with this and with Lpal carrying a transposon insertion that could excise in germinal tissue, Southern analysis showed that progeny with variegated colour derived from the crosses above were homozygous for an Lpal allele (Lpal-1) whereas all their fully red sibs tested (six from the five families) were heterozygous for Lpal-1 and putative reversion alleles (Fig. 2B).
(iii) Lpal-1 carries a CACTA transposon (Tl1, Transposon Linaria 1) in the 5′-UTR
Transposons of two classes, distinguishable according to the sequences of their terminal inverted repeats (TIR) and sizes of their target site duplications (TSD), give rise to unstable alleles in Antirrhinum. One class includes Tam3 (the same family as Ac/Ds in maize; Hehl et al., Reference Hehl, Nacken, Krause, Saedler and Sommer1991) and the other class includes a number of transposons, such as Tam1, which together with other plant transposons (e.g. Spm in maize) belong to a family known as CACTA, after a common sequence motif at their TIRs (Nacken et al., Reference Nacken, Piotowiak, Saedler and Sommer1991; Kunze & Weil, Reference Kunze, Weil and Craig2002). To determine whether Lpal-1 carries a transposon of one of these classes we amplified DNA from Lpal-1 homozygotes using primers for LPAL and for the ends of either Tam3 or Tam1. This resulted in amplification of DNA with a Tam1 primer. Full characterization of the Lpal-1 allele then showed that it carries a CACTA transposon (Tl1, Transposon Linaria 1) in the 5′UTR, 44 bp upstream of the translation initiation codon and 29 bp downstream of the transcription start site (Fig. 2C). Similarity searches showed that the 342 bp sequence of the left end of Tl1 has over 60% identity with Tam1, Tam2 and Tam4, all Antirrhinum CACTA transposons.
To test further whether Tl1 was responsible for Lpal-1 instability, we took advantage of the property of imprecise transposon excision (Saedler & Nevers, Reference Saedler and Nevers1985; Coen et al., Reference Coen, Carpenter and Martin1986, Reference Coen, Robbins, Almeida, Hudson, Carpenter, Berg and Howe1989). Duplications of short sequences at host loci generated on integration (TSD) can be retained intact or in diverse modified forms following excision. In addition, transposon excision can generate deletions adjacent to the TSD. As is typical for CACTA transposons, Tl1 was flanked by a 3 bp TSD in Lpal-1 (the trinucleotide TAA, which had a single copy in plants with red flowers of unknown relatedness to the variegated plants; Fig. 2D). In germinal reversion alleles and in single large somatic red clones from which DNA could be extracted we found no cases with a single TAA sequence, six independent cases in which both copies of the TSD were retained intact, three cases with modified TSDs and one case in which a 10 bp deletion was generated adjacent to the TSD (Fig. 2D). Presumably, these small sequence modifications did not have a large effect on LPAL expression given their location in the 5′UTR.
(iv) Concluding remarks
In conclusion, by applying knowledge derived from the analysis of unstable pigmentation mutants in Antirrhinum to a case of flower colour variegation in Linaria we have trapped an active copy of a transposon belonging to the CACTA family in the gene encoding dihydroflavonol-4-reductase. This may open the way for developing a system to select for transposon insertions in Linaria counterparts of Antirrhinum genes such as, for example, those controlling flower morphology. The two genera have readily comparable flowers with similar structures and overall dorsoventral patterns (Cubas et al., Reference Cubas, Vincent and Coen1999; Gübitz et al., Reference Gübitz, Caldwell and Hudson2003). However, they differ in details of petal size and shape as well as in some conspicuous traits. The most obvious is a long spur in the ventral petal of Linaria which is thought to be the natural counterpart of ventral petal outgrowths arising in some gain-of-function Antirrhinum mutants (Golz et al., Reference Golz, Keck and Hudson2002). Thus, Antirrhinum and Linaria might provide a useful pair of close relatives for investigating origins of both subtle and large morphological variations.